Most enterprises have an API problem they cannot put a number on. Teams fragment, integrations multiply, and every new consumer (frontend, partner, AI agent) becomes its own project. The assessment answers one question for engineering and digital leadership. Should we standardise on a federated supergraph, and if we do, what does it unlock?
"everyone has a graphql strategy. few have federation in production."
Engineering teams are under pressure to ship faster, plug in AI agents, and onboard new partner channels without rewriting the stack each time. Most cannot.
Integration projects do not compound. Vendor pitches do not answer the architecture question. Internal debates run for quarters without converging. The gap between organisations that have moved and organisations still planning is widening by the quarter.
01
Every new frontend, partner channel, or AI consumer needs its own integration work. Leadership knows the cost is real. No one has put a number on it, so no one can build a credible case for changing it.
02
Apollo, WunderGraph, or The Guild. Federation, schema stitching, or REST scaled up. Vendor pitches do not give you an answer you can defend internally. Your team has argued for three quarters without converging.
03
AI agents need typed, governed access to backend systems. Without a coherent API layer underneath, every agent integration becomes a security review your team cannot pass.
five days. one leadership-ready document at the end. the assessment answers the architecture question in writing.
Shaped to land on a leadership desk. Engineering detail sits in a separate appendix.
Architecture maturity rating across five dimensions, with a defensible scoring rationale per dimension.
Prioritised use-case shortlist mapped to measurable time-to-market or integration cost reduction.
Independent vendor recommendation across Apollo, WunderGraph and The Guild, with reasoning specific to your stack.
90-day action plan with named owners and decision points.
Executive summary in business language. Engineering appendix in technical detail. Two audiences, one engagement.
years delivering GraphQL Federation in production
largest retailer running on one supergraph (Rensa Family, 7 brands)
major federation platforms with production experience
days from kick-off to a roadmap your CTO can act on
Five years ago, federation was a Netflix pattern. Gartner now classifies it as the governance architecture for multi-team API estates.
AI that acts. Booking, ordering, automating operational work. Most enterprise pilots stall on the security review of the API surface they want to expose.
The protocol AI systems speak to your systems. Anthropic released it in late 2024 and handed it to the Linux Foundation in 2025. Production usage at Apollo, AWS, Microsoft and Bloomberg.
From first conversation to a clear, prioritised roadmap
Intake questionnaire plus three focused interviews with tech, architecture and commercial stakeholders.
Preliminary maturity assessment synthesised from the interviews, with a workshop agenda seeded with real findings.
3-4 hour facilitated session with 4 to 8 cross-functional participants. The agenda comes in prepared. The outputs are decided in the room.
The full deliverable set: maturity rating, use-case shortlist, vendor recommendation, action plan, executive summary.
We present findings and the roadmap to the broader stakeholder group. Follow-on opportunities surface naturally.
Whoever runs agents safely against their supergraph ships customer-facing features faster and automates operational work that stays manual elsewhere. The gap widens by the quarter.
Every hand-written integration produces a fragile authorisation layer. What starts as a quick pilot turns into a maintenance backlog. Federation-first removes that whole cost line.
Organisations folding GraphQL into an existing replatform get federation for almost no extra cost. Waiting means doing the work twice.
The assessment is designed for mid-to-large organisations with a multi-team API estate and a senior decision-maker who can sponsor the outcomes. Deep GraphQL expertise is not required. Willingness to look honestly at where you stand, and to act on what the assessment surfaces, is.
Typical triggers. A stalled GraphQL initiative. A replatform from a monolithic stack. A partner channel growing faster than your integration team can keep up with. An AI initiative that cannot get past security review.
have decided against GraphQL on principle and want their position validated
want a vendor comparison and call that strategy
want tactical implementation without engaging on architecture or governance
are pure greenfield startups where a single GraphQL schema is fine
Federation sits at the core of Evolve, our composable delivery practice. We hold production experience with Apollo GraphOS, WunderGraph Cosmo, and The Guild's Hive, which makes us credible as an independent adviser. No vendor reseller margin to defend.
Our open-source toolkit graphql-codegen-mcp-tools is the only production-grade pattern we know of for exposing a supergraph safely to AI agents. The engineers who wrote it deliver the assessment.
common questions
A central GraphQL API works fine for a single team. The moment you have multiple backend teams contributing to one customer-facing schema, the central layer becomes the bottleneck. Federation lets each team own their slice of the schema (a subgraph) and the platform composes them into one supergraph. Teams move independently. Consumers see one coherent API.
No. Federation is a layer that sits on top of your existing stack. REST APIs, SOAP services, custom integrations and legacy backends stay where they are. The assessment tells you which parts to keep, which to refactor, and which to leave alone.
Whichever fits your context. We have production experience with Apollo GraphOS, WunderGraph Cosmo and The Guild's Hive. No reseller margin to defend. The independence is what you pay for.
Most of this work pays back without agents in the picture. Federation improves delivery speed, governance, and team autonomy on its own. Agent readiness is an option you can take or leave once the foundation is in place.
€15K ex. VAT for the five-day assessment. The output is a maturity rating, a vendor recommendation, a 90-day action plan, and a leadership-ready executive summary. Most clients use it to decide whether to engage us, another partner, or their own team for the follow-on work. The assessment is useful either way.
Call us to talk through how our GraphQL services can help your business. Get ahead of the API pressure, and get ready for AI.